I have a machine whose SSD contains two ext4 partitions (approx 70 GiB and 40 GiB) and no EFI partition.
I want to use the space occupied by the 70 GiB partition to install KDE Linux - i.e. I expect the installer to make two partitions in that space (a 4 GiB EFI and a 66 GiB btrfs). To install, I’m booting into a USB drive to which I wrote the kde-linux_202602240255.raw image.
However, when I choose “Replace partition” and ask the installer to replace the 70 GiB partition, it wants to use that whole 70 GiB for the partition, and not create an EFI partition. It wants to use the existing EFI partition at /dev/sdb1, but that won’t work - /dev/sdb1 is the EFI partition of the installer USB drive, it doesn’t live on a fixed drive of the system.
The behaviour is similar in “Install alongside” mode. If I try to shrink a partition to leave space, then the installer wants to use all the newly-freed space for the btrfs partition, because again it thinks it can use the installer USB’s EFI partition instead of creating a new one.
It looks like the “search first for an existing EFI partition that you can use” logic needs to be made more selective so that it ignores the installer USB device.
There’s a related but not identical Invent issue here. In that case, the user has another fixed drive with an EFI partition, and the installer offers the choice of either the installer USB’s EFI partition, or the fixed drive EFI partition (which is not unreasonable, but the user isn’t given the option to create a new EFI partition on the drive that KDE Linux will be installed to).
Let me know if I should raise a new ticket in Bugzilla or Invent for this.